THE    L-XYLULOSE-XYLITOL      ENZYME     AND                    OTHER        POLYOL
               DEHYDROGENASES      OF GUINEA                      PIG
                     LIVER  MITOCHONDRIA*
            (From the Department of Biochemistry, Vandwbilt    University
                     School of Medicine, Nashville, Tennessee)

                      (Received for publication,    July 30, 1969)

   The mitochondria     of guinea pig liver contain an enzymatic system
which can reduce L-xylulose, the characteristic sugar in the urine of pento-

                                                                                             Downloaded from by guest, on March 26, 2013
suric individuals, to xylitol (2). In view of the possible importance of
this system in the normal metabolism of the ketopentose, further investi-
gation has been made of the transformation and of its possible relationship
to other enzymatic reactions. We reported briefly (3) that a preparation
obtained from the insoluble portion of ruptured guinea pig liver mitochon-
dria contains at least two distinct polyol dehydrogenases (“ketose reduc-
tases”) : (1) a triphosphopyridine   nucleotide (TPN)-dependent      enzyme
which catalyzes the interconversion of n-xylulose and xylitol, and (2) a
diphosphopyridine     nucleotide (DPN)dependent     system which acts on
n-xylulose and xylitol as well as on several other ketoses and polyols.
The present paper describes the preparation of these enzymes in soluble
form and characterizes them in regard to substrate specificity, stability,
and other properties.


   Materials-L-Xylulose       was prepared by isomerization     of L-xylose in
pyridine (2). n-Xylulose was obtained from n-xylose by a similar pro-
cedure . Xylitol was prepared by refluxing n-xylose for 7 hours in 166
parts of 70 per cent ethanol containing 10 parts of Raney nickel (aged 5
months) (4). The filtered solution was evaporated, under vacuum, to a
green syrup which was extracted with boiling absolute ethanol.       Evapora-
tion of the extract yielded a light yellow syrup which crystallized from
ethanol.     Three recrystallizations    from ethanol and, finally, one from
methanol gave a total yield (including fractions from mother liquors) of
36 per cent, m.p. 91.5-92.5’.        Dr. N. K. Richtmyer, of the National In-
   * This study was supported by a grant from the National Science Foundation.           A
preliminary   account wac presented before the Forty-seventh     annual meeting of the
American Society of Biological Chemists at Atlantic City, April, 1956 (1).
   t Fulbright Scholar, 19554%; aided by a grant from the United States Department
of State. Permanent address, Phyeiologisch-chemiechea       Institut   der Universitlit,
Gattingen, Germany.
88                      L-XYLULOSE-XYLITOL      ENZYME

stitute of Arthritis and Metabolic Diseases,generously furnished n-threitol,
n-arabitol, n-talitol, n-gulitol, n-iditol, D-glycero-D-mannoheptitol (volemi-
tol) , D-gly Cero-D-glUcOheptitO1   @-sedoheptitol) , L-glycero-n-mannoheptitol
 (perseitol), D-sorbose, n-iditol hexaacetate, and cY-D-altroheptulose hexa-
acetate. The D-iditol derivative was catalytically deacetylated at room
temperature with barium methoxide. Since in this procedure a good
crystalline product was not obtained, paper chromatographic assay was
employed to estimate the amount of the desired polyol in the preparations
tested. The deacetylation of the altroheptulose derivative was carried
 out with barium methoxide at 0”. Paper chromatography indicated that
 a high yield of sedoheptulose was obtained.
    The following substrates were kindly contributed by other investigators:

                                                                                     Downloaded from by guest, on March 26, 2013
 n-ribulose (as the o-nitrophenylhydrazone),       Dr. B. L. Horecker, of the
National Institute of Arthritis and Metabolic Diseases; L-ribulose, Dr.
W. A. Wood and Dr. F. Simpson, of the University of Illinois; n-fuculose (as
 the o-nitrophenylhydrazone), Dr. M. Greene and Dr. S. S. Cohen, of the
 Children’s Hospital of Philadelphia; L-erythrulose, Dr. G. C. Mueller, of
 the University of Wisconsin; L-mannitol, Dr. E. Baer, of the Banting
 Institute.    All other substrates were commercial samples. Erythritol,
 L-arabitol, and dulcitol were obtained from the Mann Research Labora-
 tories, Inc.; n-mannitol (mannite), from Merck and Company, Inc.; ribitol
 (adonitol), from the Nutritional Biochemicals Corporation; n-sorbitol, from
 Matheson, Coleman, and Bell; n-fructose, from the Difco Laboratories,
 Inc.; n-sorbose, from the Pfanstiehl Chemical Company; allitol, n-manno-
 heptulose, and n-glucoheptulose, from General Biochemicals, Inc.
     DPN, DPNH, and TPN were products of the Pabst Laboratories and
 of the Sigma Chemical Company. These coenzymes were at least 90 per
 cent pure. Muscle lactic dehydrogenase was purchased from the Worth-
 ington Biochemical Corporation, and spinach chloroplast TPN- diaphorase
 was a gift from Dr. Andre Jagendorf of the McCollum-Pratt             Institute,
 The Johns Hopkins University.          According to Dr. Jagendorf, 1 unit of
 enzyme is the amount causing a change in optical density of 1.0 Beckman
 unit per minute at 620 mp with 2,3,6-trichlorophenol-indophenol            as the
 electron acceptor. He advised us that 1 unit per 3.0 ml. of reaction mix-
 ture would be an ample excess of enzyme.
     Methods-All       spectrophotometric measurements were made with a
 Beckman model DU spectrophotometer. Unless otherwise indicated, sub-
  strate specificity tests were carried out at 23” in quartz cuvettes having
 approximately 3 ml. capacity and a 1.00 cm. light path. Contents of a
  typical test solution (3.0 ml. total volume) were as follows: 0.3 ml. of 0.5
  M tris(hydroxymethyl)aminomethane         (Tris) buffer at pH 8.1, 0.3 ml. of
  0.08 M MgCh, 0.3 ml. of enzyme extract, 0.392 pmole of coenzyme, and
                        S.   HOLLMANN   AND   0.   TOUSTER                      89

19.7 pmoles of polyol or 6.66 pmoles of ketose. When several substances
were tested during an experiment, they were usually added in 0.05 ml. of
water in order to minimize changes in concentration of components already
present. Total volumes of 1 ml. were used for a few substrates available
in very limited amounts, all components being reduced proportionately.
Readings were made against solutions containing all components except
    Color reactions used for identification      of ketoses were (1) the cysteine-
carbazole reaction (5)) (2) the orcinol reaction (6), with a 40 minute heating
period, and (3) a cysteine-sulfuric acid reaction kindly furnished by Dr. G.
Ashwell of the National Institute of Arthritis and Metabolic Diseases.’
Before carrying out the various calorimetric analyses, reaction mixtures

                                                                                      Downloaded from by guest, on March 26, 2013
were deproteinized by addition of 2 volumes each of 5 per cent ZnSO1.-
7HzO and 0.29 N Ba(OH)2 and of 5 volumes of water, the reagents having
been adjusted in strength to yield a final pH of approximately          7.2.
    Preparation of Enzyme-The          entire procedure was carried out at O-2”
except when otherwise indicated.          The liver mitochondria of fasted guinea
pigs (Carworth Farms, Inc.) were prepared as described previously (2),
except that the mitochondrial        pellet was washed three times with 0.15 M
KCl-0.01 M NaHC03.            For the mitochondria     from 30 to 40 gm. of liver,
the volumes of these successive washes were 180 ml., 120 ml., and, finally,
a volume equal to 53 per cent of the weight of the liver used (yielding a
suspension equivalent to a “65 per cent homogenate” of the liver). After
the last centrifugation      (10 minutes, 3400 r.p.m., International     PR-1 cen-
trifuge, rotor No. 823), the pellet was suspended in the “53 per cent”
volume of water. The suspension was allowed to stand for 1 hour, and
then the rupturing of the mitochondria was completed with a motor-driven
homogenizer composed of a Teflon pestle and smooth glass tube. Mito-
chondria could no longer be detected by phase microscopy.            The insoluble
portion of the ruptured particles was collected by centrifugation at 10,000
r.p.m. for 10 minutes in the high speed attachment (rotor No. 296) of the
International    centrifuge.     The precipitate (the “mitochondrial      residue”)
was at times tested directly, in which case it was suspended in the 53 per
cent volume of water. In other experiments the precipitate was washed
with water or with phosphate buffer prior to testing its activity.
    To liberate the dehydrogenases, the residue was subjected to a modifica-
tion of Morton’s butanol procedure for the isolation of succinic dehydro-
genase (7). In a typical experiment, the mitochondrial            residue from 43
 gm. of liver was suspended in 25 ml. of 0.02 M sodium phosphate buffer at
pH 7.88, the total volume being 33.5 ml. The suspension was cooled to
 -4’to     -5’, and 13.4 ml. (40 per cent of the volume of the residue suspen-
   l To be published.
90                        L-XYLULOSE-XYLITOL             ENZYME

sion) of n-butyl alcohol were slowly added, with continuous stirring, from a
capillary burette.      The addition was carried out over a period of at least 30
minutes.     An addition time of 1 to 2 hours seems to enhance recovery of
most of the dehydrogenases,        but a specific study of this variable was not
made. The mixture was stirred for 30 minutes after completion of the
butanol addition.        The emulsion was centrifuged at 19,000 r.p.m. for 30
minutes, the butanol layer was decanted, the mat of denatured protein was
 removed with a spatula, and the clear, yellow aqueous layer filtered into
 a flask for lyophilization.    The primary purpose of the lyophilization    is to
 remove the butanol, since dialysis inactivates         the TPN enzyme.       The
 dried product can be dissolved in water and filtered to yield a solution which
is stable for many weeks in the refrigerator.

                                                                                            Downloaded from by guest, on March 26, 2013
   Properties of Mitochmdrial       Residue-Practically      all of the activity of
the mitochondria      towards L-xylulose is present in the insoluble portion
of the particles, the supernatant solution after centrifugation         having negli-
gible action on the pentose.      However, if the residue iswashed with water,
it is no longer active in the presence of the usual constituents of the incuba-
tion medium, namely, phosphate buffer, MgCh, and glutamate.                    (Adeno-
sine triphosphate     was also added because it had been required in early
mitochondrial     experiments    (2).)   On the assumption        that the washing
process removes soluble glutamic dehydrogenase               which adheres to the
precipitate, DPNH was substituted for the glutamate.               Fig. 1 shows that
this replacement led to utilization of the ketose.            The interpretation       of
this experiment is not clear. Addition of the soluble mitochondrial             portion
plus glutamate did not permit xylulose utilization by the washed residue
or by enzyme rendered soluble.          Furthermore,     as will be evident below,
relatively little of the L-xylulose enzyme can be obtained in soluble form
if the residue is washed with water preliminary          to the butanol treatment.
Nonetheless, the experiments with washed residue and DPNH led to em-
phasis on the pyridine nucleotide requirement             and therefore facilitated
initial studies on the soluble extract.
    Some experiments on the mitochondrial           residue (unwashed)       indicated
that (1) Tris buffer and phosphate buffer are equally useful, (2) the optimal
pH for L-xylulose disappearance is approximately           7.5, and (3) magnesium
 chloride and glutamate are required for maximal activity.                 The mito-
 chondrial residue could convert xylitol to ketopentose to a small extent
 (5.8 per cent) if DPN and methylene blue were added to the usual medium
at pH 7.42. However, increasing the alkalinity was more effective than
 addition of coenzyme and dye in reversing the reaction.               A 15 per cent
yield was obtained in 90 minutes in a flask containing 0.1 ml. of 0.5 M Tris
                            S.     HOT&MANN   AND   0.   TOUSTER                             91

buffer at pH 9.0,O.l ml. of 0.08 M MgCh,            20.7 pmoles of DPN, 19.9 ccmoles
of xylitol, and 0.2 ml. of a suspension of           mitochondrial residue (in 0.02 M
phosphate buffer at pH 7.9) equivalent              to a 50 per cent homogenate of
whole liver in a total volume of 1.0 ml.            The ketopentose was determined

                                                                   0 SOL. EXTRACT   f DPNH

                                                                                                  Downloaded from by guest, on March 26, 2013
               TIME   IN MINUTES                                   TIME   IN MINUTES
                        Fxa. 1                                    FIQ. 2
    FIG. 1. Utilization     of L-xyluloae by insoluble portion of ruptured mitocbondria.
Each flask contained 0.3 ml. of 0.5 M glutamate, 0.9 ml. of 0.154 M phosphate buffer
at pH 7.4, 0.3 ml. of Tris buder at pH 7.60.3 ml. of 0.08 M MgCL, 0.6 ml. of a sus-
pension of residue in the 53 per cent volume of water, and 26 pmoles of xylulose.
ATP ‘was added as 0.3 ml. of a 0.06 M solution; total volume, 3.0 ml. The flask in the
DPNH trial contained 26.6 gmoles of NazDPNH and 13.3 &moles of L-xylulose in a
 total volume of 2.0 ml., all other components being reduced proportionately.            The
washed residue was prepared by twice suspending a portion of a residue preparation
in the 53 per cent volume of water and collecting it each time by centrifugation           for
 10 minutes at 16,OW r.p.m. Analysis of Ba-Zn filtrates by the cysteine-carbazole
    FIG. 2. Utilization     of L-xylulose by enzyme solution obtained from mitochondrial
residues. The flask with residue contained 0.3 ml. of 0.5 M glutamate, 0.9 ml. of
0.154 M phosphate buffer at pH 7.4, 0.3 ml. of 0.03 M MgCL, 0.6 ml. of residue in the
53 per cent volume of water, 0.3 ml. of the soluble portion of ruptured mitochondria,
 and 20 pmoles of xylulose; total volume, 3.0 ml. (The soluble mitochondrial        fraction
 was later found to be unnecessary.)        The extract waz tested in a solution containing
 0.3 ml. of 0.154 M phosphate buffer at pH 7.4,0.1 ml. of 0.03 M MgCl2, 13.3 pmoles of
 NazDPNH, 0.2 ml. of enzyme extract, and 6.66 pmoles of xylulose; total volume, 1.0
 ml. Analysis of Ba-Zn filtrates by the cysteine-carbazole        method.

in Ba-Zn filtrates by the cysteine-carbazole method. The residue had no
action on n-erythrulose, n-sorbose, or n-arabitol.
   Soluble n-Xylulose-Xylitol System-Fig. 2 shows that extraction of the
n-xylulose-reducing enzyme is accomplished in high yield. It should
be noted that the ratio of DPNH to n-xylulose in this experiment was 2: 1.
Early in the work with the soluble enzyme solution it was found that, if
the coenzyme is limiting (xylitol to coenzyme ratio of 50: l), TPN is much
92                              L-XYLULOSE-XYLITOL                  ENZYME

more effective than DPN in promoting the dehydrogenation of xylitol.            It
was tentatively assumed that the enzyme resembled other infrequently
encountered enzymes which are not very specific in their coenzyme require-
ment. However, tests of ketose reduction in the presence of limiting
amounts of the reduced coenzymes indicated clearly that they act in dif-
ferent enzymatic reactions. DPN is involved in the interconversion of
xylitol and n-xylulose and of several other polyols and their corresponding
ketoses. TPN, on the other hand, is the coenzyme in the interconversion
of xylitol and L-xylulose by an enzyme which has unique substrate speci-
ficity., With TPN (or TPNH for ketoses), the following compounds are
practically inactive as substrates: L-threitol, erythritol, ribitol, D-arabitol,
L-arabitol, D-sorbitol, dulcitol, n-mannitol, L-mannitol, D-talitol, n-gulitol,

                                                                                                             Downloaded from by guest, on March 26, 2013
D-idikd,   L-iditO1,   ditO1,           D-glyCerO-D-~~OheptitOl,                     D-&CerO-D-glUCO-

                                    0    40      00       120  160 200       240
                                                TIME     IN MINUTES
  Fro. 3. Substrate specificity          test of TPN           system.   Procedure      as described under
“Methods”   (3 ml. cells).
heptitol, L-glycero-n-mannoheptitol,     L-erythrulose, D-ribulose, L-ribulose,
n-xylulose, n-fructose, n-sorbose, L-sorbose, L-fuculose, sedoheptulose,
n-mannoheptulose,      and n-glucoheptulose.      Only the n-gulitol could be
considered to have slight activity; this indication may have been due to
a trace of impurity.     A typical substrate specificity test is illustrated in
Fig, 3. The influence of pH on the equilibrium point is shown by the ef-
fects of the HCl additions in lowering the final concentration of TPNH.
Except for L-xylulose, no ketose caused a decrease in absorption at 340
rnp. The identification of the xylitol product as xylulose has already been
reported (3), the evidence having been based upon absorption spectra of
the colors produced in the orcinol and in the cysteine-carbazole reactions
and on the rate of color formation in the latter test.
   Efforts were made to increase the yield of ketose. The first experiment
    f We suggest that this enzyme be named I%-xylulose (xylitol) reductase.”    The
 existence of the two xylitol enzyme systems reported in this paper, as well as our
 recent finding (unpublished)    of L-arabitol, perhaps derived from L-xylulose, in
 pentosuric urine, makes it desirable to include both the ketose and polyol in the
                          S.    H.OLLMANN   AND   0.   TOTJSTER             93

involved the use of borate, but 20 pmoles of this agent, added at equilib-
rium to the regular incubation mixture, increased the TPNH concentra-
tion by only a small amount.       Even a solution containing Tris buffer at
pH 9.0, MgCL, enzyme extract, and a xylitol to TPN ratio of 1: 1.3 gave
a yield of only 2.7 per cent (by cysteine-carbazole analysis after Ba-Zn
deproteinization).    Dr. Jagendorf’s TPN-diaphorase         was then used in
an attempt to increase the yield. One experiment with 2,6dichlorophenol-
indophenol was unsuccessful, but methylene blue proved useful. For one
methylene blue trial the initial composition of the incubation mixture was
as follows: 0.50 ml. of 0.5 M Tris buffer at pH 8.12,0.50 ml. of 0.08 M MgCh,
1.0 ml. of guinea pig enzyme solution, 329 moles of xylitol, 2 units of
TPN-diaphorase,      16.44 pmoles of TPN, and 20.5 pmoles of methylene

                                                                                  Downloaded from by guest, on March 26, 2013
blue in a total volume of 2.8 ml. Two further additions of enzymes and
water were made during the 25 hour incubation at 21-23’ (with occasional
shaking of the flask). The 6nal amounts were for guinea pig enzyme, 2.0
ml.; diaphorase, 6 units; total volume, 5.0 ml. The yield of ketopentose
was 5.8 per cent. A similar experiment involved the use of 12 units of the
diaphorase, 32.86 pmoles of TPN, and an incubation for 28 hours at 24-26”.
The yield was 7.2 per cent.
   It has been shown that equal amounts of xylulose and TPNH are formed
in a solution containing buffer, MgCh, xylitol, and TPN (3). It was
therefore possible to determine, without directly measuring the amount of
xylulose formed, the equilibrium     constant for the reaction
                 Xylitd        + TPN+ S L-xylulose + TPNH + H+
The enzyme preparation used in determining the constant had been heated
at 50’ for 20 minutes, a treatment which usually causes a slight increase in
activity.   With incubation mixtures initially containing 22.97 pmoles of
xylitol, 0.4576 Ltmole of TPN, 0.35 ml. of 0.5 M Tris buffer, 0.35 ml. of
0.08 M MgCh, and 0.35 ml. of heated enzyme solution in a total volume of
3.50 ml., and with final pH values of four flasks varying from 6.96 to 8.70,
an average K of 8.58 X lo+ M was obtained after the 4 hours required for
equilibrium   to be established.
   Other Properties of TPN System-An experiment has been reported which
shows that little L-xylulose enzyme is obtained if the mitochondrial    residue
is washed three times with water prior to the butanol treatment (3). Ap-
proximately 40 per cent of the activity is lost if the residue is washed twice
with 0.02 M phosphate buffer at pH 7.88 prior to extraction of enzyme.
This is in contrast to the effect on the DPN-dependent          enzymes, which
are increased in yield (or activity) if the residue is first washed with either
water or phosphate (see below).
   Another property of the L-xylulose enzyme by which it can be contrasted
with the other dehydrogenases is its ready inactivation by dialysis.      Reac-
tion between TPN and xylitol could not be detected with enzyme which
94                       L-XYLULoSE-XYLITOL       ENZYME

had been dialyzed against water (3 ml. of extract, 2 liters of water at 2”,
 16 hours of dialysis).       Even with use of the procedure of Nicholas and
Nason (8), in which the dialysis membrane is first soaked in 0.15 M phos-
phate at pH 7.4 containing lo-’ M glutathione and the dialysis is then
carried out against 0.10 M phosphate containing 5 X 10-* M cysteine, only
37 per cent of the activity of undialyzed enzyme was obtained.            The dial-
ysis time was 16 hours.
    In spite of the lability of the enzyme to dialysis and to washing of the
mitochondrial      residue, it is stable in aqueous solution for many weeks at
2’ and for 20 minutes at 50”. It is stable at pH 4 for 2 hours at 2’ and, as
many long spectrophotometric           experiments show clearly, at pH 9 for sev-
eral hours at room temperature.

                                                                                        Downloaded from by guest, on March 26, 2013
    The xylitol-L-xylulose     interconversion catalyzed by the soluble enzyme
preparation is almost completely inhibited by 0.004 M iodoacetate, a finding
in accord with the results obtained with intact mitochondria (2). Amino-
pyrine, a weak inhibitor of the mitochondrial          system (2), had no effect at
a concentration of 0.02 M. Although magnesium chloride was always in-
cluded in the medium, the question of its essentiality is still undecided.
Magnesium ions were clearly necessary for full activity of particulate prep-
arations.     With enzyme rendered soluble, magnesium chloride had a small
stimulatory     effect and 0.002 M sodium versenate caused only a little in-
hibition.    0.003 M dinitrophenol       was without effect.
    Substrates DPN-Dependent Tranqfomatims- Evidence for the exist-
ence of more than one mitochondrial            polyol dehydrogenase was obtained
when the enantiomorphic          forms of xylulose were tested in incubation mix-
tures containing DPNH and TPNH formed by the dehydrogenation                       of
xylitol.    With DPNH, L-xylulose had little effect, but n-xylulose caused a
rapid decrease in concentration of the reduced coenzyme. Studies men-
tioned above showed clearly that the DPN and TPN enzyme systems dif-
fered in their response to various treatments.’           The substrate specificity
of the DPN system was then studied with the view towards obtaining a
more complete characterization, especially in relation to the “soluble” (non-
particulate) liver polyol dehydrogenase of Blakley (9), which catalyzes the
interconversion of xylitol and D-xylulose (10). Table I summarizes these
substrate specificity tests. The typical behavior of several active sub-
strates is shown in Fig. 4, together with the inhibitory effect of 0.004 M
iodoacetate.      The lack of substrate activity of L-xylulose, L-sorbose, and
D-fructose (all used in 6.66 eole amounts) and the rapid reduction of
D-xylulose and of n-ribulose were shown directly in experiments with crys-
  8 The DPN and TPN systems have recently       been separated   from each other (FL
Wohl and 0. Touster, unpublished results).
                                    S. HOLLMANN                AND         0.     TOUSTER                                        95

talline DPNH.                  In these experiments, readings were made against solu-
tions containing               substrate but no coenzyme. Corrections were made for
                                                         TABLE         I
                                             &&i&f/          of DPN             &/Ste??Z

                                                                                                        Indicated     reaction
         Substrate               eactivitj              Substrate                      leactivity

n-Threitol                           +            n-Erythrulose                            +        L-Threitol       G= L-eryth-
Xylitol                                           D-Xylulose                               +        Xylitol       * D-xylulose

                                                                                                                                      Downloaded from by guest, on March 26, 2013
Ribitol                              :            D-Ribulose                               +        Ribitol      ti D-ribulose
D-Arsbitol                                        n-Ribulose
n-Arabitol                                        n-Xylulose
D-Sorbitol                           +            D-Fructose                               4-*      D-Sorbitol        C= D-fruc-
D-Iditolt                                         D-Sorbose
n-Iditol                             +            t-Sorbose                                -I-*     L-Iditol        C= L-sorbose
Dulcitol                             -
D-Mannitol                           -
D-Talitol                            -
D-Glycero-D-gluco-                   +            Sedoheptuloset                                    D-Glycero-D-glucO-
  heptitol                                                                                            heptitol c= L-gulo-
                                                                                                      heptulose (?)$
n-Glycero-D-man-                                  D-Mannoheptulose
D-Glycero-D-man-                                  D-Glucoheptuloset

    * Activity could be detected only when used at concentrations      24 times that used
with other active ketoses (see under “Methods”         and “Hexitol-ketohexose      trans-
    t 1.0 ml, reaction mixtures were used in these cases. All other substances were
tested in standard 3 ml. procedure described under “Methods.”
    $ It is possible that sedoheptulose is the ketose reactant and that, owing to an
especially low equilibrium    constant, a very high concentration   would be required to
show its activity.

the slow decrease in DPNH                            concentration                which occurred in the absence
of a reactive ketose.
    X@itoCD-X&lose                    Interwnversion -Our                         previous report (3) presented
evidence that the dehydrogenation                                   product         of xylitol in the presence of
96                        L-XYLULOSE-XYLITOL             ENZYME

DPN is xylulose and that equal amounts of the ketopentose and reduced
coenzyme are produced.       Equilibrium     constants at various pH values were
determined as described above for the xylitol-L-xylulose          transformation,
except that DPN was substituted          for the TPN and that five flasks with
final pH values varying from 7.01 to 8.76 were employed.            An average K
of 1.55 X lo-l1 M was obtained.
   Efforts to increase the yield of ketopentose were based on coupling of
the transformation    to pyruvate with lactic dehydrogenase.         Two different
batches of the polyol dehydrogenase gave xylulose yields of 2.5 and 5.9 per
cent, respectively.    In the latter experiment the incubation mixture had
the following initial composition: 0.5 ml. of 0.5 M Tris buffer at pH 8.12,
0.5 ml. of 0.08 M MgCh, 1.0 ml. of guinea pig enzyme solution, 329 pmoles

                                                                                           Downloaded from by guest, on March 26, 2013
                                          TIME   IN MINUTES
     FIQ. 4. Activity of DPN system in interconversion    of ribitol and n-ribulose and
of xylitol and n-xylulose; inhibition  of ketopentose reduction by iodoacetate.     Pro-
cedure as described under “Methods”        (3 ml. cells). The sodium iodoacetate was
added as 0.22 ml. of a 0.06 M solution.

of xylitol, 0.2 ml. of a 1:250 dilution of Worthington lactic dehydrogenase
in 0.01 M NaCl, 32.86 pmoles of DPN, and 300 pmoles of sodium pyruvate
in a total volume of 3.0 ml. During the 28 hour incubation at 24-26”
(with occasional shaking of the flask), two 0.5 ml. additions of guinea pig
enzyme and five 0.2 ml. additions of lactic dehydrogenase solution were
made. In the former experiment, in which the lower yield was obtained,
only half these amounts of lactic dehydrogenase and DPN were employed.
   The xylitol-n-xylulose enzyme system was usually activated somewhat
by preliminary washing of the mitochondrial residue (see below) and by
dialysis. Like the L-xylulose enzyme, it is stable between pH 4 and 9 and
is inhibited by 0.004 M iodoacetate (Fig. 4).
   Rib&l-n-Ribulose TransforrnatimL-It was of interest to characterize the
product of the DPN-ribitol reaction. A reaction mixture similar to the
one employed with xylitol (3) was used, 90 mg. of charcoal (Darco G-60)
                          8.        HOLLMANN           AND      0.     TOUSTER             97

per 1.8 ml. of Ba-Zn filtrate being required to yield a DPN-free      solution
for the orcinol determination.    The peaks of the absorption spectrum of
the color produced had the following ratios: 540:670 nn~, 0.75; 435~670
mp, 0.63. Authentic ribulose gave ratios of 0.75 and 0.66, respectively.
Both the ribitol product and ribulose showed the same rapid attainment
of maximal color in the cysteine-carbazole        test (less than 10 minutes).
Coupling of the ribitol dehydrogenation   to lactic dehydrogenase and sodium
pyruvate, with the same procedure as was used with xylitol, gave keto-
pentose yields of 3.2 and 5.4 per cent, the comparable values for xylitol
being 2.5 and 5.9 per cent, respectively.      The similar yields of xylulose
and ribulose from the corresponding pentitols might suggest that the same

                                                                                                Downloaded from by guest, on March 26, 2013
                      0        40       80       120        0     20     40      60   00
                                               TIME    IN    MINUTES
   FIG. 5. Reaction  of DPNH with relatively high concentrations    of n-fructose and
L-sorbose. Procedure aa described under “Methods”      (3 ml. cells). Unless other-
wise indicated, the amounts of polyols and ketoses added were 19.7 and 6.66 pmoles,

enzyme was catalyzing the two reactions. However, not all enzyme
preparations had similar ratios of activity towards the two pentitols. With
all enzyme preparations, xylitol and n-xylulose reacted more rapidly than
did ribitol and n-ribulose, respectively.   Fig. 4 shows a comparison of the
two pentitol reactions and the inhibition of n-ribulose reduction by iodo-
   Hex&d-Ketohexose Transforma2ionsThese reactions presented a prob-
lem because the usual spectrophotometric assay procedure failed to show
activity of the ketohexoses expected to be produced from n-sorbitol and
from L-iditol (Fig. 5, A). n-sorbitol was an active substrate of all enzyme
preparations; yet n-fructose and L-sorbose,the two possibleproducts (assum-
ing dehydrogenation of the second or fifth alcohol grouping), showed no
activity under the usual very favorable conditions for ketopentose reduc-
98                            L-XYLULOSE-xmIT0L     mznm

tion.    L-Sorbose is also the expected product from n-iditol.        It was there-
fore necessary to characterize the dehydrogenation            products by various
calorimetric and paper chromatographio          methods.
    That fructose is indeed produced from D-sorbitol is indicated by the fol-
lowing evidence.      Firstly, the sorbitol product and fructose gave similar
 colors in the cysteine-carbazole     test (565 rnp peak), in the orcinol reaction
 (Fig. S), and in the cysteine-sulfuric     acid test (Fig. 6). The colors in the
latter two tests were unlike the colors given by sorbose (Fig. 7). Secondly,
paper chromatographic         comparison, with 80 per cent n-propyl alcohol as
solvent and naphthoresorcinol        spray (ll), showed identical Rp values and

                                                                                          Downloaded from by guest, on March 26, 2013
            WAVE     LENGTH   (mp)                         WAVE     LENGTH   (mp)

                   FIG.   6                                       FIG.   7
   FIG. 6. Absorption  spectra of h-fructose and of the u-sorbitol product in color
tests. 0, u-sorbitol product; 0, u-fructose, in the orcinol reaction; A, u-sorbitol
product; A, D-fructose, in the cysteine-sulfuric     acid reaction.
    FIG. 7. Absorption spectra of L-sorbose and of the L-iditol product in color tests.
 l , L-iditol product; 0, L-sorbose, in the orcinol reaction; A, L-iditol product;
A, L-sorbose, in the cysteine-sulfuric  acid reaction.

color development. Thirdly, inclusion of n-fructose in two incubation
mixtures caused a decrease in rate of DPNH formation, from sorbitol and
DPN, proportionate to t,he fructose concentration. Finally, by use of a
very high concentration of n-fructose (24 times the effective ketopentose
level), it was possible to show that the ketohexose can effect the slow oxida-
tion of DPNH (Fig. 5, B).
   The L-iditol reaction was studied similarly. Fig. 7 shows that the reac-
tion product and sorboseyield similar colors in the orcinol reaction and in
the cysteine-sulfuric acid test. In the cysteine-carbazole test, identical
spectra and rates of color development were obtained. Furthermore, pa-
per chromatographic comparison (80 per cent n-propyl alcohol solvent and
naphthoresorcinol spray) provided additional confirmation. As with
n-fructose, the use of large amounts of n-sorbosefinally disclosed its activity
                                      S.   HOLLMANK            AND        0.     TOUSTER                                            99

as a substrate (Fig. 5, B). The inactivity of a similar concentration of
n-sorbose is shown for comparison.
   Multiplicity of DPN-Dependent    Polyol Dehydrogenases-Evidence     for the
presence of more than one DPN-dependent       polyol dehydrogenase in mito-
chondria was provided by the observation that dehydrogenation rates for
several polyols varied unpredictably with various enzyme extracts. While
our earlier extracts showed greater activity towards xylitol than towards
all other polyols, more recent ones have been at least as active towards
n-sorbitol as towards xylitol.   Comparison of polyol oxidation rates of sev-
eral extracts is shown in Table II.
   Apparent similarities among the DPN enzymes may be due to secondary
factors. Fig. 8 shows that extracts prepared from mitochondrial      residues

                                                                                                                                         Downloaded from by guest, on March 26, 2013
which have been washed with phosphate buffer have enhanced activity

                                                           TABLE     II
     Relative      Rates     of Reaction     of Xylitol,       D-Sorbitol,            and   L-Iditol          in DPN    System*

                                                                                      Enzyme    preparation
                                                            16C                If A             18A               19A         20A
Xylitol..........................                          0.024           0.048               0.029            0.024       0.040
D-Sorbitol..                                               0.010           0.016               0.034            0.039       0.045
L-Iditol      ..                                                                                                0.022       0.057

   * The values in the table are the increases in absorption at 340 rnp, during the 10
minute period following addition of substrate, of 3 ml. test solutions prepared aa
described under “Methods.”

(with DPN) towards xylitol and D-sorbitol, although the TPN-xylitol       reac-
tion rate is decreased. It is possible that a factor involved in DPN de-
struction is removed in the washing process (or by dialysis, which usually
activates the DPN-xylitol   reaction).     It may be mentioned here that wash-
ing of the mitochondrial  residue with water also activates the DPN-xylitol
system although the TPN-xylitol      reaction is decreased even more markedly
than when phosphate is used (3).


   It is necessary to emphasize that, in all the work prior to the studies on
the soluble preparations, attention was paid almost exclusively to activity
towards n-xylulose.     In fact, under the ordinary test conditions, particu-
late preparations showed no activity towards any compounds other than
n-xylulose and xylitol.    Even n-xylulose, a very active substrate with the
aqueous extract, was not acted upon by mitochondria         under conditions
100                          L-XYLULOSE-XYLITOL             ENZYME

which led to reduction of L-xylulose (2). It is therefore possible that polyol
dehydrogenases other than the ones reported are present in mitochondria
but have not been examined appropriately,      perhaps being inactivated or
removed at some step in the fractionation.     The lack of activity of D-xy-
lulose with mitochondria   may be due to the association of the xylitol-n-
xylulose enzyme with an inhibitor which is removed during the prepara-
tion of extract. The increased yield of the n-xylulose enzyme from washed
mitochondria and the activation of extracts by dialysis provide some sup
port for this explanation.

                                                                                         Downloaded from by guest, on March 26, 2013
                        20     40    0     20    49     0     20   40
                                    TIME     IN MINUTES
    FIQ. 8. Change in soluble polyol dehydrogenase activity resulting from washing
of mitochondrial   residues with phosphate buffer. A preparation      of residue was
divided into two portions.    One was suspended in phosphate buffer, and butanol
was added as described under “Preparation       of enzyme.”   The other portion was
suspended twice in the proportionate     volume of phosphate buffer and sedimented
each time by centrifugation    for 10 minutes at 16,000 r.p.m. The precipitate     was
then resuspended in buffer and subjected to the butanol procedure.     l , extract ob-
tained from unwashed residue; 0, extract obtained from washed residue.

   Although our DPN system differs in cellular location from that of Blak-
ley’s enzyme, similarities in substrate specificity (10) led to concern about
the possible identity of the two systems. Three differences in substrate
reactivity are now evident: (1) n-fructose and L-sorbose are much less
reactive with our extract than with Blakley’s rat liver extract; (2) sedo-
heptulose, very active with a purified preparation from the soluble fraction
of sheep liver (lo), is inactive with the extract derived from guinea pig
liver mitochondria;     (3) allitol, another substrate for the Blakley prepara-
tion (lo), is inactive with ours. In retrospect it is realized that the ques-
tion of identity of the mitochondrial        and non-mitochondrial    systems is
probably a meaningless one, since it appears likely that both are mixtures
of closely related enzymes. Evidence for the multiplicity       of DPNdepend-
ent polyol dehydrogenases in guinea pig liver mitochondria           has already
been presented in this paper. Williams-Ashman          and Banks (12), who re-
                           5.   HOLLMANN     AND   0.   TOUSTER                          101

ported that the Blakley enzyme occurs also in the seminal vesicle and in
the coagulating gland of the rat, found that L-arabitol is a substrate for a
rat liver extract.    This is in disagreement with another study (10). Fur-
ther work is obviously necessary on the purification and characterization
of the various liver polyol dehydrogenases.
    The equilibrium constants of both of the xylitol-xylulose                reactions are
considerably smaller than those reported for the sorbitol-n-fructose                   reac-
tion catalyzed by t.he non-mitochondrial           liver enzyme (9, 12) and for the
mannitol-1-phosphate-n-fructose-6-phosphate             reaction catalyzed by an en-
zyme extracted from Escherichiu coli B (13).
    The guinea pig was originally chosen for the urinary L-xylulose studies
because it resembled man in requiring L-ascorbic acid in the diet. Then,

                                                                                                Downloaded from by guest, on March 26, 2013
the liver of bhe guinea pig was used for the enzyme studies because the keto-
pentose was found to be a constituent of its urine (14). These considera-
tions may actually have had little influence on the progress of this investi-
gation, but our few experiments on other species suggest that the Guinea
pig was a fort,unat,e choice. Hamster, rat, and monkey liver preparations
utilized L-xylulose more slowly than comparable ones from guinea pigs.
    The physiological significance of the mitochondrial              dehydrogenases       re-
mains to be elucidated.        Enzymatic activity of the kind reported here has
not .been previously detected in mitochondria.             Since the DPN system is
unquestionably     most active towards n-xylulose among the ketoses tested,
and is also quite active towards xylitol, it is possible that the successive
action of the two xylulose enzymes, both present in the same morphological
unit, effects the interconversion       of the enantiomorphic       forms of the ketose.
There is considerable evidence that n-glucuronolactone                    is a metabolic
precursor of L-xylulose (14), but a normal role for the former substance has
not been reported.        Nevertheless,     a substance derived from a glucuronic
acid-containing    metabolite may be the normal precursor of the pentose.
    The discovery of the key role of n-xylulose-5-phosphate            in the 6-phospho-
gluconate pathway suggests a route for L-xylulose to enter (or be formed
from) a normal, major carbohydrate             pathway.       Yeast transketolase,        re-
ported recently to act on n-xylulose-5-phosphate           rather than on D-ribulose-
5-phosphate, has some action on free n-xyluiose                (15), but transketolase
isolated from spinach is without action on the free pentose (16). A specific
n-xylulose kinase from bacteria has been found which converts the pentose
to its 5-phosphate derivative        (17). The presence of a similar enzyme in
liver would provide a bridge bet.ween the xylulose transformations                  and the
6-phosphogluconate       pathway.       Studies to detect such a kinase in liver have
been initiated.4
   4 While the present paper was in proof, Hickman         and Ashaell   (18) reported   the
occurrence of n-xylulokinase in calf liver.
102                              L-XYLULOSE-XYLITOL          ENZYME

   We acknowledge the able assistance of Mrs. Ruth Hutcheson Mayberry
and Miss Margarita   Escobedo G. in the preparation    of xylulose, of Mr.
John C. Kirschman in the preparation of xylitol, and of Mr. Richard Wohl
in carrying out some substrate specificity tests.

   Studies on polyol dehydrogenases        of the insoluble portion of ruptured
guinea pig liver mitochondria       are reported.    The enzymes, made soluble
by a modification of the butanol method, show activity towards a number
of polyols and ketoses.       Data presented indicate the presence of an en-
zyme, unique among such dehydrogenases             in its triphosphopyridine      nu-
cleotide (TPN) requirement,        which catalyzes the interconversion       only of

                                                                                         Downloaded from by guest, on March 26, 2013
xylitol and L-xylulose.      The extract also contains diphosphopyridine          nu-
cleotide (DPN)-requiring       dehydrogenases which catalyze the interconver-
sion of several substrates,     namely, xylitol and n-xylulose, L-threitol and
n-er$thrulose,   ribitol and n-ribulose, n-sorbitol and n-fructose, n-iditol and
L-sorbose, and D-glycero-D-glucoheptitol      and a heptulose.     Evidence is pre-
sented that the TPN- and DPN-dependent            enzymes are distinct from each
other, and the indications for the multiplicity          of DPN-dependent       liver
polyol dehydrogenases       are pointed out. The possibility that the mito-
chondrial enzymes provide a connection between L-xylulose and the 6-
phosphogluconate       pathway is dis,cussed.

 1. Touster, O., Reynolds, V. H., Hutcheson, R. M., and Hollmann, S., Federation
      PTOC., 16, 372 (1956).
 2. Touster, O., Reynolds, V. H., and Hut,cheson, R. M., J. Biol. Chem., 221, 697
 3. Hollmenn, S., and Touster, O., J. Am. Chem. Sot., 78,3544 (1956).
 4. Karabinos, J. V., and Ballun, A. T., J. Am. Chem. Sot., 76, 4591 (1953).
 5. Dische, Z., and Borenfreund, E., J. Biol. Chem., 192, 583 (1951).
 6. Mejbaum, W., 2. physiol.    Chem., 258, 117 (1939).
 7. Morton, R. K., in Colowick, S. P., and Kaplan, N. O., Methods in enzymology,
      New York, 1, 48 (1955).
 8. Nicholas, D. J. D., and Nason, A., J. Biol. Chem., 207,353 (1954).
 9. Blakley, R. L., Biochem. J., 49, 257 (1951).
10. McCorkindale,    J., and Edson, N. L., Biochem. J., 67,518 (1954).
11. Bryson, J. L., and Mitchell, T. S., Nature, 187, 864 (1951).
12. Williams-Ashman,     H. G., and Banks, J., Arch. Biochem.    and Biophys., 50, 513
13. Wolff, J. B., and Kaplan, N. O., J. BioZ. C&m.,     218, 849 (1956).
14. Touster, O., Hutcheson, R. M., and Rice, I,., J. BioZ. Chem., 216, 677 (1955).
15. Srere, P. A., Cooper, J. R., Klybas, V., and Racker, E., Arch. Biochem. and Bio-
      phys.,   69, 535 (1955).
16. Smyrniotis, P. Z., and Horecker, B. L., J. BioZ. Chem., 218, 745 (1956).
17. Stumpf, P. K., and Horecker, B. L., J. BioZ. Chem., 218,753 (1956).
18. Hickman, J., and Ashwell, G., J. Am. Chem. Sot., 78, 6209 (1956).

To top